Field notes.
Engineering posts, changelogs, and a running log of what we're learning about branching live machines.
Exploring ProgramBench
Taking MemPalace to 100%
We improved MemPalace's retrieval logic to 100% in LongMemEval's recall@5 metric using just Python.
Bringing Hermes to WebAssembly
We compiled Hermes agent to WebAssembly with Pyodide and py2wasm then benchmarked both.
A coding agent with direction
An overview of zagent, a harness for getting more done with a single shot.
How we sped up bun by 100x
A deep dive into rewriting git in Zig and integrating it into Bun for a 100x speedup
Automate web dev with headless browser testing
Making a poker app with Elixir/Phoenix using agents testing for UX improvements
Conducting a Headless Symphony
Why and how to deploy a headless Symphony with coding agents running in Vers VMs
I spent a billion tokens and all I got was this repo
A journey through a billion tokens to bridge Elixir and WebAssembly
Captain's Log — February 11, 2026
Day 3. The one where the platform went down and we built around it.
Introducing kips
Keep your personal data across sessions with MCP clients like Claude Desktop
Introducing mcp-shell
An additional utility while we enter the next era
Psychotransport
Nolita and Index: rearchitected. Additional products in the works
Into the Wormhole
Dimensionality means pursuing lines of flight
Recap: GrowthHax
The aftermath of 60 hackers in New York City
2024-07 Release Notes
Nolita 2.1: Claude 3.5 support, replay sessions, unified auth
Memory Index Launch
Get started with integration, with more features on the way
Why agents fail
Learn the iron law
2024-06 Release Notes
Nolita 2.0: introducing the natural language, drop-in Page API
2024-01 Release Notes
Life after collective memory
About High Dimensional Research
We create infrastructure for intuition.